Mmtrack

MMTracking 1.0.0rc1 is the 2-nd version of MMTracking 1.x, a part of the OpenMMLab 2.0 projects.

Built upon the new [training engine](https://github.com/open-mmlab/mmengine), MMTracking 1.x unifies the interfaces of datasets, models, evaluation, and visualization.

And there are some BC-breaking changes. Please check [the migration tutorial](https://mmtracking.readthedocs.io/en/1.x/migration.html) for more details.

We also support more methods in MMTracking 1.x, such as StrongSORT for MOT, Mask2Former for VIS, PrDiMP for SOT.

We recommend you to use MMTracking v1.0.0rc1 version, since the v1.0.0rc0 version has some bugs about the limitation of minimum version of mmdet.

Highlights

- Introduce the 1.0.0rc0 version of MMTracking ([725](https://github.com/open-mmlab/mmtracking/pull/725))

New Features

- Support OC-SORT method for MOT ([545](https://github.com/open-mmlab/mmtracking/pull/545))

- Support multi-class tracking in ByteTrack ([548](https://github.com/open-mmlab/mmtracking/pull/548))

- Support DanceTrack dataset for MOT ([543](https://github.com/open-mmlab/mmtracking/pull/543))

- Support TAO dataset for QDTrack ([585](https://github.com/open-mmlab/mmtracking/pull/585))

Highlights

- Support tracking colab tutorial ([511](https://github.com/open-mmlab/mmtracking/pull/511))

New Features

- Refactor the training datasets of SiamRPN++ ([496](https://github.com/open-mmlab/mmtracking/pull/496)), ([#518](https://github.com/open-mmlab/mmtracking/pull/518))

- Support loading data from ceph for SOT datasets ([494](https://github.com/open-mmlab/mmtracking/pull/494))

- Support loading data from ceph for MOT challenge dataset ([517](https://github.com/open-mmlab/mmtracking/pull/517))

- Support evaluation metric for VIS task ([501](https://github.com/open-mmlab/mmtracking/pull/501))

Bug Fixes

- Fix a bug in the LaSOT datasets and update the pretrained models of STARK ([483](https://github.com/open-mmlab/mmtracking/pull/483)), ([#503](https://github.com/open-mmlab/mmtracking/pull/503))

- Fix a bug in the format_results function of VIS task ([504](https://github.com/open-mmlab/mmtracking/pull/504))

Highlights

- Support QDTrack algorithm in MOT ([433](https://github.com/open-mmlab/mmtracking/pull/433)), ([#451](https://github.com/open-mmlab/mmtracking/pull/451)), ([#461](https://github.com/open-mmlab/mmtracking/pull/461)), ([#469](https://github.com/open-mmlab/mmtracking/pull/469))

Bug Fixes

- Support empty tensor for selsa aggregator ([463](https://github.com/open-mmlab/mmtracking/pull/463))

Highlights

- Support STARK algorithm in SOT ([443](https://github.com/open-mmlab/mmtracking/pull/443)), ([#440](https://github.com/open-mmlab/mmtracking/pull/440)), ([#434](https://github.com/open-mmlab/mmtracking/pull/434)), ([#438](https://github.com/open-mmlab/mmtracking/pull/438)), ([#435](https://github.com/open-mmlab/mmtracking/pull/435)), ([#426](https://github.com/open-mmlab/mmtracking/pull/426))

- Support HOTA evaluation metrics for MOT ([417](https://github.com/open-mmlab/mmtracking/pull/417))

New Features

- Support TAO dataset in MOT ([415](https://github.com/open-mmlab/mmtracking/pull/415))